module/zfs/zpl_inode.c

   1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright (c) 2011, Lawrence Livermore National Security, LLC.
  23  */
  24
  25
  26 #include <sys/zfs_vfsops.h>
  27 #include <sys/zfs_vnops.h>
  28 #include <sys/zfs_znode.h>
  29 #include <sys/vfs.h>
  30 #include <sys/zpl.h>
  31
  32
  33 static struct dentry *
  34 #ifdef HAVE_LOOKUP_NAMEIDATA
  35 zpl_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
  36 #else
  37 zpl_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
  38 #endif
  39 {
  40         cred_t *cr = CRED();
  41         struct inode *ip;
  42         int error;
  43         fstrans_cookie_t cookie;
  44
  45         if (dlen(dentry) > ZFS_MAXNAMELEN)
  46                 return (ERR_PTR(-ENAMETOOLONG));
  47
  48         crhold(cr);
  49         cookie = spl_fstrans_mark();
  50         error = -zfs_lookup(dir, dname(dentry), &ip, 0, cr, NULL, NULL);
  51         spl_fstrans_unmark(cookie);
  52         ASSERT3S(error, <=, 0);
  53         crfree(cr);
  54
  55         spin_lock(&dentry->d_lock);
  56         dentry->d_time = jiffies;
  57 #ifndef HAVE_S_D_OP
  58         d_set_d_op(dentry, &zpl_dentry_operations);
  59 #endif /* HAVE_S_D_OP */
  60         spin_unlock(&dentry->d_lock);
  61
  62         if (error) {
  63                 if (error == -ENOENT)
  64                         return (d_splice_alias(NULL, dentry));
  65                 else
  66                         return (ERR_PTR(error));
  67         }
  68
  69         return (d_splice_alias(ip, dentry));
  70 }
  71
  72 void
  73 zpl_vap_init(vattr_t *vap, struct inode *dir, zpl_umode_t mode, cred_t *cr)
  74 {
  75         vap->va_mask = ATTR_MODE;
  76         vap->va_mode = mode;
  77         vap->va_uid = crgetfsuid(cr);
  78
  79         if (dir && dir->i_mode & S_ISGID) {
  80                 vap->va_gid = KGID_TO_SGID(dir->i_gid);
  81                 if (S_ISDIR(mode))
  82                         vap->va_mode |= S_ISGID;
  83         } else {
  84                 vap->va_gid = crgetfsgid(cr);
  85         }
  86 }
  87
  88 static int
  89 #ifdef HAVE_CREATE_NAMEIDATA
  90 zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
  91     struct nameidata *nd)
  92 #else
  93 zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
  94     bool flag)
  95 #endif
  96 {
  97         cred_t *cr = CRED();
  98         struct inode *ip;
  99         vattr_t *vap;
 100         int error;
 101         fstrans_cookie_t cookie;
 102
 103         crhold(cr);
 104         vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
 105         zpl_vap_init(vap, dir, mode, cr);
 106
 107         cookie = spl_fstrans_mark();
 108         error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
 109         if (error == 0) {
 110                 VERIFY0(zpl_xattr_security_init(ip, dir, &dentry->d_name));
 111                 VERIFY0(zpl_init_acl(ip, dir));
 112                 d_instantiate(dentry, ip);
 113         }
 114
 115         spl_fstrans_unmark(cookie);
 116         kmem_free(vap, sizeof (vattr_t));
 117         crfree(cr);
 118         ASSERT3S(error, <=, 0);
 119
 120         return (error);
 121 }
 122
 123 static int
 124 zpl_mknod(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
 125     dev_t rdev)
 126 {
 127         cred_t *cr = CRED();
 128         struct inode *ip;
 129         vattr_t *vap;
 130         int error;
 131         fstrans_cookie_t cookie;
 132
 133         /*
 134          * We currently expect Linux to supply rdev=0 for all sockets
 135          * and fifos, but we want to know if this behavior ever changes.
 136          */
 137         if (S_ISSOCK(mode) || S_ISFIFO(mode))
 138                 ASSERT(rdev == 0);
 139
 140         crhold(cr);
 141         vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
 142         zpl_vap_init(vap, dir, mode, cr);
 143         vap->va_rdev = rdev;
 144
 145         cookie = spl_fstrans_mark();
 146         error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
 147         if (error == 0) {
 148                 VERIFY0(zpl_xattr_security_init(ip, dir, &dentry->d_name));
 149                 VERIFY0(zpl_init_acl(ip, dir));
 150                 d_instantiate(dentry, ip);
 151         }
 152
 153         spl_fstrans_unmark(cookie);
 154         kmem_free(vap, sizeof (vattr_t));
 155         crfree(cr);
 156         ASSERT3S(error, <=, 0);
 157
 158         return (error);
 159 }
 160
 161 static int
 162 zpl_unlink(struct inode *dir, struct dentry *dentry)
 163 {
 164         cred_t *cr = CRED();
 165         int error;
 166         fstrans_cookie_t cookie;
 167
 168         crhold(cr);
 169         cookie = spl_fstrans_mark();
 170         error = -zfs_remove(dir, dname(dentry), cr);
 171         spl_fstrans_unmark(cookie);
 172         crfree(cr);
 173         ASSERT3S(error, <=, 0);
 174
 175         return (error);
 176 }
 177
 178 static int
 179 zpl_mkdir(struct inode *dir, struct dentry *dentry, zpl_umode_t mode)
 180 {
 181         cred_t *cr = CRED();
 182         vattr_t *vap;
 183         struct inode *ip;
 184         int error;
 185         fstrans_cookie_t cookie;
 186
 187         crhold(cr);
 188         vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
 189         zpl_vap_init(vap, dir, mode | S_IFDIR, cr);
 190
 191         cookie = spl_fstrans_mark();
 192         error = -zfs_mkdir(dir, dname(dentry), vap, &ip, cr, 0, NULL);
 193         if (error == 0) {
 194                 VERIFY0(zpl_xattr_security_init(ip, dir, &dentry->d_name));
 195                 VERIFY0(zpl_init_acl(ip, dir));
 196                 d_instantiate(dentry, ip);
 197         }
 198
 199         spl_fstrans_unmark(cookie);
 200         kmem_free(vap, sizeof (vattr_t));
 201         crfree(cr);
 202         ASSERT3S(error, <=, 0);
 203
 204         return (error);
 205 }
 206
 207 static int
 208 zpl_rmdir(struct inode * dir, struct dentry *dentry)
 209 {
 210         cred_t *cr = CRED();
 211         int error;
 212         fstrans_cookie_t cookie;
 213
 214         crhold(cr);
 215         cookie = spl_fstrans_mark();
 216         error = -zfs_rmdir(dir, dname(dentry), NULL, cr, 0);
 217         spl_fstrans_unmark(cookie);
 218         crfree(cr);
 219         ASSERT3S(error, <=, 0);
 220
 221         return (error);
 222 }
 223
 224 static int
 225 zpl_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 226 {
 227         boolean_t issnap = ITOZSB(dentry->d_inode)->z_issnap;
 228         int error;
 229         fstrans_cookie_t cookie;
 230
 231         /*
 232          * Ensure MNT_SHRINKABLE is set on snapshots to ensure they are
 233          * unmounted automatically with the parent file system.  This
 234          * is done on the first getattr because it's not easy to get the
 235          * vfsmount structure at mount time.  This call path is explicitly
 236          * marked unlikely to avoid any performance impact.  FWIW, ext4
 237          * resorts to a similar trick for sysadmin convenience.
 238          */
 239         if (unlikely(issnap && !(mnt->mnt_flags & MNT_SHRINKABLE)))
 240                 mnt->mnt_flags |= MNT_SHRINKABLE;
 241
 242         cookie = spl_fstrans_mark();
 243         error = -zfs_getattr_fast(dentry->d_inode, stat);
 244         spl_fstrans_unmark(cookie);
 245         ASSERT3S(error, <=, 0);
 246
 247         return (error);
 248 }
 249
 250 static int
 251 zpl_setattr(struct dentry *dentry, struct iattr *ia)
 252 {
 253         struct inode *ip = dentry->d_inode;
 254         cred_t *cr = CRED();
 255         vattr_t *vap;
 256         int error;
 257         fstrans_cookie_t cookie;
 258
 259         error = inode_change_ok(ip, ia);
 260         if (error)
 261                 return (error);
 262
 263         crhold(cr);
 264         vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
 265         vap->va_mask = ia->ia_valid & ATTR_IATTR_MASK;
 266         vap->va_mode = ia->ia_mode;
 267         vap->va_uid = KUID_TO_SUID(ia->ia_uid);
 268         vap->va_gid = KGID_TO_SGID(ia->ia_gid);
 269         vap->va_size = ia->ia_size;
 270         vap->va_atime = ia->ia_atime;
 271         vap->va_mtime = ia->ia_mtime;
 272         vap->va_ctime = ia->ia_ctime;
 273
 274         cookie = spl_fstrans_mark();
 275         error = -zfs_setattr(ip, vap, 0, cr);
 276         if (!error && (ia->ia_valid & ATTR_MODE))
 277                 error = zpl_chmod_acl(ip);
 278
 279         spl_fstrans_unmark(cookie);
 280         kmem_free(vap, sizeof (vattr_t));
 281         crfree(cr);
 282         ASSERT3S(error, <=, 0);
 283
 284         return (error);
 285 }
 286
 287 static int
 288 zpl_rename(struct inode *sdip, struct dentry *sdentry,
 289     struct inode *tdip, struct dentry *tdentry)
 290 {
 291         cred_t *cr = CRED();
 292         int error;
 293         fstrans_cookie_t cookie;
 294
 295         crhold(cr);
 296         cookie = spl_fstrans_mark();
 297         error = -zfs_rename(sdip, dname(sdentry), tdip, dname(tdentry), cr, 0);
 298         spl_fstrans_unmark(cookie);
 299         crfree(cr);
 300         ASSERT3S(error, <=, 0);
 301
 302         return (error);
 303 }
 304
 305 static int
 306 zpl_symlink(struct inode *dir, struct dentry *dentry, const char *name)
 307 {
 308         cred_t *cr = CRED();
 309         vattr_t *vap;
 310         struct inode *ip;
 311         int error;
 312         fstrans_cookie_t cookie;
 313
 314         crhold(cr);
 315         vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
 316         zpl_vap_init(vap, dir, S_IFLNK | S_IRWXUGO, cr);
 317
 318         cookie = spl_fstrans_mark();
 319         error = -zfs_symlink(dir, dname(dentry), vap, (char *)name, &ip, cr, 0);
 320         if (error == 0) {
 321                 VERIFY0(zpl_xattr_security_init(ip, dir, &dentry->d_name));
 322                 d_instantiate(dentry, ip);
 323         }
 324
 325         spl_fstrans_unmark(cookie);
 326         kmem_free(vap, sizeof (vattr_t));
 327         crfree(cr);
 328         ASSERT3S(error, <=, 0);
 329
 330         return (error);
 331 }
 332
 333 static void *
 334 zpl_follow_link(struct dentry *dentry, struct nameidata *nd)
 335 {
 336         cred_t *cr = CRED();
 337         struct inode *ip = dentry->d_inode;
 338         struct iovec iov;
 339         uio_t uio;
 340         char *link;
 341         int error;
 342         fstrans_cookie_t cookie;
 343
 344         crhold(cr);
 345
 346         iov.iov_len = MAXPATHLEN;
 347         iov.iov_base = link = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
 348
 349         uio.uio_iov = &iov;
 350         uio.uio_iovcnt = 1;
 351         uio.uio_resid = (MAXPATHLEN - 1);
 352         uio.uio_segflg = UIO_SYSSPACE;
 353
 354         cookie = spl_fstrans_mark();
 355         error = -zfs_readlink(ip, &uio, cr);
 356         spl_fstrans_unmark(cookie);
 357         if (error) {
 358                 kmem_free(link, MAXPATHLEN);
 359                 nd_set_link(nd, ERR_PTR(error));
 360         } else {
 361                 nd_set_link(nd, link);
 362         }
 363
 364         crfree(cr);
 365         return (NULL);
 366 }
 367
 368 static void
 369 zpl_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr)
 370 {
 371         const char *link = nd_get_link(nd);
 372
 373         if (!IS_ERR(link))
 374                 kmem_free(link, MAXPATHLEN);
 375 }
 376
 377 static int
 378 zpl_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
 379 {
 380         cred_t *cr = CRED();
 381         struct inode *ip = old_dentry->d_inode;
 382         int error;
 383         fstrans_cookie_t cookie;
 384
 385         if (ip->i_nlink >= ZFS_LINK_MAX)
 386                 return (-EMLINK);
 387
 388         crhold(cr);
 389         ip->i_ctime = CURRENT_TIME_SEC;
 390         igrab(ip); /* Use ihold() if available */
 391
 392         cookie = spl_fstrans_mark();
 393         error = -zfs_link(dir, ip, dname(dentry), cr);
 394         if (error) {
 395                 iput(ip);
 396                 goto out;
 397         }
 398
 399         d_instantiate(dentry, ip);
 400 out:
 401         spl_fstrans_unmark(cookie);
 402         crfree(cr);
 403         ASSERT3S(error, <=, 0);
 404
 405         return (error);
 406 }
 407
 408 #ifdef HAVE_INODE_TRUNCATE_RANGE
 409 static void
 410 zpl_truncate_range(struct inode *ip, loff_t start, loff_t end)
 411 {
 412         cred_t *cr = CRED();
 413         flock64_t bf;
 414         fstrans_cookie_t cookie;
 415
 416         ASSERT3S(start, <=, end);
 417
 418         /*
 419          * zfs_freesp() will interpret (len == 0) as meaning "truncate until
 420          * the end of the file". We don't want that.
 421          */
 422         if (start == end)
 423                 return;
 424
 425         crhold(cr);
 426
 427         bf.l_type = F_WRLCK;
 428         bf.l_whence = 0;
 429         bf.l_start = start;
 430         bf.l_len = end - start;
 431         bf.l_pid = 0;
 432         cookie = spl_fstrans_mark();
 433         zfs_space(ip, F_FREESP, &bf, FWRITE, start, cr);
 434         spl_fstrans_unmark(cookie);
 435
 436         crfree(cr);
 437 }
 438 #endif /* HAVE_INODE_TRUNCATE_RANGE */
 439
 440 #ifdef HAVE_INODE_FALLOCATE
 441 static long
 442 zpl_fallocate(struct inode *ip, int mode, loff_t offset, loff_t len)
 443 {
 444         return (zpl_fallocate_common(ip, mode, offset, len));
 445 }
 446 #endif /* HAVE_INODE_FALLOCATE */
 447
 448 static int
 449 #ifdef HAVE_D_REVALIDATE_NAMEIDATA
 450 zpl_revalidate(struct dentry *dentry, struct nameidata *nd)
 451 {
 452         unsigned int flags = (nd ? nd->flags : 0);
 453 #else
 454 zpl_revalidate(struct dentry *dentry, unsigned int flags)
 455 {
 456 #endif /* HAVE_D_REVALIDATE_NAMEIDATA */
 457         zfs_sb_t *zsb = dentry->d_sb->s_fs_info;
 458         int error;
 459
 460         if (flags & LOOKUP_RCU)
 461                 return (-ECHILD);
 462
 463         /*
 464          * After a rollback negative dentries created before the rollback
 465          * time must be invalidated.  Otherwise they can obscure files which
 466          * are only present in the rolled back dataset.
 467          */
 468         if (dentry->d_inode == NULL) {
 469                 spin_lock(&dentry->d_lock);
 470                 error = time_before(dentry->d_time, zsb->z_rollback_time);
 471                 spin_unlock(&dentry->d_lock);
 472
 473                 if (error)
 474                         return (0);
 475         }
 476
 477         /*
 478          * The dentry may reference a stale inode if a mounted file system
 479          * was rolled back to a point in time where the object didn't exist.
 480          */
 481         if (dentry->d_inode && ITOZ(dentry->d_inode)->z_is_stale)
 482                 return (0);
 483
 484         return (1);
 485 }
 486
 487 const struct inode_operations zpl_inode_operations = {
 488         .create         = zpl_create,
 489         .link           = zpl_link,
 490         .unlink         = zpl_unlink,
 491         .symlink        = zpl_symlink,
 492         .mkdir          = zpl_mkdir,
 493         .rmdir          = zpl_rmdir,
 494         .mknod          = zpl_mknod,
 495         .rename         = zpl_rename,
 496         .setattr        = zpl_setattr,
 497         .getattr        = zpl_getattr,
 498         .setxattr       = generic_setxattr,
 499         .getxattr       = generic_getxattr,
 500         .removexattr    = generic_removexattr,
 501         .listxattr      = zpl_xattr_list,
 502 #ifdef HAVE_INODE_TRUNCATE_RANGE
 503         .truncate_range = zpl_truncate_range,
 504 #endif /* HAVE_INODE_TRUNCATE_RANGE */
 505 #ifdef HAVE_INODE_FALLOCATE
 506         .fallocate      = zpl_fallocate,
 507 #endif /* HAVE_INODE_FALLOCATE */
 508 #if defined(CONFIG_FS_POSIX_ACL)
 509 #if defined(HAVE_GET_ACL)
 510         .get_acl        = zpl_get_acl,
 511 #elif defined(HAVE_CHECK_ACL)
 512         .check_acl      = zpl_check_acl,
 513 #elif defined(HAVE_PERMISSION)
 514         .permission     = zpl_permission,
 515 #endif /* HAVE_GET_ACL | HAVE_CHECK_ACL | HAVE_PERMISSION */
 516 #endif /* CONFIG_FS_POSIX_ACL */
 517 };
 518
 519 const struct inode_operations zpl_dir_inode_operations = {
 520         .create         = zpl_create,
 521         .lookup         = zpl_lookup,
 522         .link           = zpl_link,
 523         .unlink         = zpl_unlink,
 524         .symlink        = zpl_symlink,
 525         .mkdir          = zpl_mkdir,
 526         .rmdir          = zpl_rmdir,
 527         .mknod          = zpl_mknod,
 528         .rename         = zpl_rename,
 529         .setattr        = zpl_setattr,
 530         .getattr        = zpl_getattr,
 531         .setxattr       = generic_setxattr,
 532         .getxattr       = generic_getxattr,
 533         .removexattr    = generic_removexattr,
 534         .listxattr      = zpl_xattr_list,
 535 #if defined(CONFIG_FS_POSIX_ACL)
 536 #if defined(HAVE_GET_ACL)
 537         .get_acl        = zpl_get_acl,
 538 #elif defined(HAVE_CHECK_ACL)
 539         .check_acl      = zpl_check_acl,
 540 #elif defined(HAVE_PERMISSION)
 541         .permission     = zpl_permission,
 542 #endif /* HAVE_GET_ACL | HAVE_CHECK_ACL | HAVE_PERMISSION */
 543 #endif /* CONFIG_FS_POSIX_ACL */
 544 };
 545
 546 const struct inode_operations zpl_symlink_inode_operations = {
 547         .readlink       = generic_readlink,
 548         .follow_link    = zpl_follow_link,
 549         .put_link       = zpl_put_link,
 550         .setattr        = zpl_setattr,
 551         .getattr        = zpl_getattr,
 552         .setxattr       = generic_setxattr,
 553         .getxattr       = generic_getxattr,
 554         .removexattr    = generic_removexattr,
 555         .listxattr      = zpl_xattr_list,
 556 };
 557
 558 const struct inode_operations zpl_special_inode_operations = {
 559         .setattr        = zpl_setattr,
 560         .getattr        = zpl_getattr,
 561         .setxattr       = generic_setxattr,
 562         .getxattr       = generic_getxattr,
 563         .removexattr    = generic_removexattr,
 564         .listxattr      = zpl_xattr_list,
 565 #if defined(CONFIG_FS_POSIX_ACL)
 566 #if defined(HAVE_GET_ACL)
 567         .get_acl        = zpl_get_acl,
 568 #elif defined(HAVE_CHECK_ACL)
 569         .check_acl      = zpl_check_acl,
 570 #elif defined(HAVE_PERMISSION)
 571         .permission     = zpl_permission,
 572 #endif /* HAVE_GET_ACL | HAVE_CHECK_ACL | HAVE_PERMISSION */
 573 #endif /* CONFIG_FS_POSIX_ACL */
 574 };
 575
 576 dentry_operations_t zpl_dentry_operations = {
 577         .d_revalidate   = zpl_revalidate,
 578 };